CN101051059A

CN101051059A - Waveform compression and display

Info

Publication number: CN101051059A
Application number: CNA2007100896207A
Authority: CN
Inventors: S·K·萨利文; P·M·格尔拉赫; K·韦思; K·P·多拜恩斯
Original assignee: Tektronix Inc
Current assignee: Tektronix Inc
Priority date: 2006-03-20
Filing date: 2007-03-20
Publication date: 2007-10-10
Also published as: JP4955432B2; EP1837665A3; US20070217694A1; EP1837665A2; JP2007256284A; US7834780B2

Abstract

A waveform compression and display technique saves both a peak detected version (background version) and a decimated/lowpass filtered version (foreground version) of a sampled electrical signal. The two versions are displayed simultaneously overlaid together in a contrasting manner so as not to obscure information contained in either of them. The lowpass filtered version uses a series of simple lowpass filters (22) with decimation to produce a single data stream from a plurality of data streams derived from the sampled electrical signal. The single data stream may then be subjected to additional filtering, such as a cascaded integrator-comb filter (100), to obtain a desired frequency bandwidth. When displayed, the peak detect pixels adjacent the decimated/lowpass filtered pixels may be adjusted in intensity so that the low frequency information of the lowpass filtered waveform is not lost, while the peak detect pixels further from the lowpass filtered pixels are intensified to highlight the high frequency information. Alternatively the background version intensity may be controlled by a user control over a first range from zero to a predetermined maximum, and the foreground version may be controlled over a second range from a default intensity to a maximum, saturated intensity.

Description

Waveform compression and demonstration

Technical field

The present invention relates to electric signal waveform measurement and demonstration, more particularly, relate to the method for waveform compression and demonstration.

Background technology

In digital oscilloscope, Wave data is by quick sampling and be stored in the storer.When observing for a long time section, data volume may surpass available storer.When this situation took place, some information need go out of use.So sampled data needed to compress before storing storer into.Because data arrive with high speed, so available compression scheme is limited.The most frequently used compression scheme is:

Extract: this compression scheme aliasing high-frequency information is also omitted burst pulse.

Peak value detects: low-frequency information is lost when having high-frequency signal.

HiRes filtering: some high-frequency informations are filtered off by aliasing and burst pulse.

Each all has important disadvantages in these schemes, as detailed below.

In digital oscilloscope, need to check high-frequency signal.In order to accomplish that in real time this point, sampling rate are necessary for the twice of the highest frequency component that is examined signal.This realizes with the following methods: use very fast fabrication techniques A-D converter (ADC), then with many this ADC placement in parallel, the same analog input channel of digitizing, its mode is for each sampling clock cycle, and each ADC digitizing is from the data of different time.Very high sampling rate is to have accomplished, but the problem of the huge incremental data that is produced in the very short time period can occur how managing.

Sampled data is written in the very fast storer usually, the many storeies of placement in parallel as required.As long as the checked time period is very short and data volume is suitable for memory depth, so just effectively.But oscilloscope user wants to check electric signal in long-time section, no matter and memory depth have muchly, data volume all can substantially exceed memory depth.In order to allow long-time section to check, need compression scheme and reduce the data volume that is stored in the storer.When checking in the very long time period, compression scheme must can have very large ratio of compression.Because data arrive from ADC with two-forty like this, so compression scheme also must be very fast.

Extraction is the simple form of compression.If compress with factor 10, discarded 9 get final product in per 10 samplings.But extracting is not an extraordinary compression scheme, because for example may very important burst pulse often go out of use.And high frequency component signal is aliased into low frequency.This can make important low-frequency information fuzzy.This also can make the compression " unloyal " of this form, because the low frequency signal of being seen may be not really there.Exercise question is the U.S. Patent No. 5 of " Anti-Aliasing DitheringMethod and Apparatus for Low Frequency Signal Sampling ", 115,189 have illustrated a kind of method that keeps stable high fdrequency component from the low frequency signal that seems aliasing, that is: when extracting, from every group of N sampling, select a sampling that will keep at random with N.High fdrequency component still is aliased into low frequency, but firm high-frequency signal is aliased into many low frequencies by this frequency modulation (PFM) of sampling rate.Exercise question is that the U.S. Patent No. 6,388,595 of " Dithering Apparatus toProperly Represent Aliased Signals for High Speed Signal Sampling " has covered identical scheme, but implementation has change.

Another change scheme that extracts is shown in exercise question in the U.S. Patent No. 4,586,022 of " Waveform Memory Circuit ", wherein sampled data and the final value that is stored in the storer is compared.When the difference between these two samplings surpasses a limit, the hits that newly is worth and be extracted is stored in the storer together.Use this technology, compare with the signal with high fdrequency component, mainly the signal of being made up of low frequency component has much higher ratio of compression.But in the general digital oscillograph, this technology is not quite effective, because all signals, even the signal with high fdrequency component, all need be compressed so that be fit to memory depth.

The peak value inspection comprises the time is divided into many equal time periods, finds out peak signal voltage and minimum signal voltage in each time period then.Originally be to use mimic channel to find out minimum and maximum voltage.With these voltage digitalizations and be stored in the storer.Recently come the check dight signal to reach identical result with the digital peak testing circuit.The simulated peak detecting device is always introduced error, and if signal is not just at peak value sampling, then digital peak value detector just can not be deposited crest voltage.When sampling rate was more much bigger than the highest frequency component of signal, this back one problem was just little.Though peak value detects a kind of " loyalty " mode that can be considered to the display simulation signal, it trends towards making signal to have noise, because it has shown peak noise.And when having high fdrequency component, they trend towards making low frequency component fuzzy.

Common minimum and maximum value all is written into storer during each time interval.But exercise question is the U.S. Patent No. 4,183,087 of " Peak Deviation Sampling " has illustrated a kind of technology, wherein only with minimum or maximal value, looks which and former selected value variation maximum, is kept in the storer.But owing to only storing the inaccuracy that a half data is introduced, this scheme generally is unworthy adopting.Nowadays be described in the U.S. Patent No. 4,271,486 that the general scheme of using is " Waveform StorageSystem " at exercise question.The minimum and maximum digital value of each time period is found out and is kept in the storer with Digital Logic.Exercise question is that the U.S. Patent No. 4,755,960 of " WaveformData Compressing Circuit " has illustrated a little change to basic peak value detection scheme.In a kind of change, each peak value detects can be by last to changing, and when mapping, a perpendicular line extends to and touches last perpendicular line like this.Other change comprises to be followed the tracks of overrange and owes the range code, all values is added a constant and allows data not carry out peak value detect and just pass through.

Exercise question is that the U.S. Patent No. 5,115,404 of " Digital Storage Oscilloscope with Indication of AliasedDisplay " has illustrated that the another kind that peak value detects changes.Except finding out the minimum and maximum value on the time period, described circuit is also found out the number of times of signal slope variation and which is followed the tracks of find earlier, is maximal value or minimum value.The sequential storage that minimum and maximum value takes place according to their is in storer.The indicator that also comprises twice of generation or more times slope variation.Data are shown as the band between the minimum and maximum value then.On the other hand, when slope do not change twice or more times the time, this data presentation is that this line of 2 is passed in maintenance in proper order.

Exercise question is the U.S. Patent No. 5 of " Method and Apparatus for Detecting/Storing Waveform PeakValue ", 547,232 have illustrated a kind of scheme, wherein data produce so rapidly, so that single maximal value and minimum value circuits built are got up very expensive, so data are divided in groups-may be parallel data stream-and find out maximal value and minimum value in every group.Exercise question is that the U.S. Patent No. 6,121,799 of " Interleaved Digital Peak Detector " has illustrated a kind of peak detection circuit, wherein data is divided into many pipelines or path, and finds out the peak value in each path.

Exercise question is that the U.S. Patent No. 5,740,064 of " Sampling Technique for Waveform Measuring Instruments " has illustrated how to obtain the blended data that peak value detects data and extracted data.When peak amplitude during greater than a limit, peak value detects data and is stored in the storer, otherwise the storage extracted data.This is rendered as rarer making an uproar with regard to the low frequency signal that allows to have noise.Though shown data seem quite a lot of, this has the effect that makes peak value detecting pattern " unloyal ".Exercise question is that the U.S. Patent No. 6,344,844 of " Digital Oscilloscope Having Improved Peak DetectMode " shows that peak value detects and extracted data all can be kept at and obtains in the storer.The data that extract are shown then according to the arbitrary condition in many conditions arranged different the demonstration but peak value detects data by routine.It should be noted that most that then peak value detection data can show with low-intensity if the vertical height of peak value detection data is very little.This has low noise low frequency with regard to allowing main use extracted data to draw, and peak noise just can not emphasized like this.But with higher-strength draw with in the past or the significantly different peak value of data subsequently detect that the ability of data such as burst pulse still exists.

In general, preventing a kind of mode of aliasing before filtering is considered to be at and extracts, is the U.S. Patent No. 4,621 of " Anti-Aliasing Filter Circuit for Oscilloscopes " as exercise question, shown in 217.This scheme prevents aliasing by removed high frequency component signal before carrying out extraction.Filtering was finished before signal is digitized.Exercise question is that the U.S. Patent No. 4,802,098 of " Digital BandpassOscilloscope " is optionally removed frequency component, so only leaves given frequency band, allows the compression in available memory on the required time section.

Exercise question is that the U.S. Patent No. 5,233,546 of " Anti-Alias Filtering Apparatus for Frequency DomainMeasurements " has illustrated a kind of technology of coming the digitized filtered data with finite impulse response (FIR) low-pass filter.In high sampling rate, in storer, obtain earlier data, filtering then.But do not realize compression.In enough low sampling rate, digitalized data can filtered and extraction before being stored in storer.Many modern oscillographs, such as Beaverton, the pattern that TDS400, TDS500B that the Tektronix company of Oregon makes and TDS700A digital oscilloscope all have a kind of being called " HiRes ".This obtaining mode uses moving average filter, extracts then, with packed data.When being used for observing low frequency signal, the HiRes wave filter has significantly reduced high frequency noise.Unfortunately, the HiRes moving average filter is not good low-pass filter.It allows some quite a large amount of high-frequency signals to pass through, and these high-frequency signals are then by aliasing and be rendered as low frequency component.

Needed is fast a kind of and intelligent manner more, comes the compression sampling Wave data to keep the demonstration of " loyalty ".

Summary of the invention

Therefore, the invention provides a kind of waveform compression and display technique, the peak value that it preserves the sampling electric signal detects version (version) (" background " version) and extraction and/or low-pass filtering version (" prospect " version).These two kinds of versions show simultaneously with way of contrast, and do not make the contained information fuzzy of each edition.The prospect version can only be used extraction, or can use a series of simple low pass filter to produce single data stream together from a plurality of data stream that the sampling electric signal is derived with extracting, and the amount of low-pass filtering can not regulated from having to the required grade of maximum.Single data stream then can be through additional filtering, to obtain required frequency span.When being shown as overlapping waveform simultaneously, the background version pixel of contiguous prospect version pixel can be regulated as a kind of contrast version on intensity, the low-frequency information of extraction/low-pass filtering waveform just can not lost like this, and can be reinforced with bright apparent high-frequency information apart from prospect version pixel background version pixel far away.By suitable identification prospect and background version pixel, just can use the sole user to be controlled at the intensity from zero to greatest level intensity adjustments background version pixel on first scope, and from the default value to the maximum saturation value, regulate the intensity of prospect version pixel.Also can show and realize contrast by different colours between use prospect and the background version pixel or saturation grade.

Purpose of the present invention, advantage and other novel feature are read from following detailed description and in conjunction with claims and accompanying drawing, may be obvious that.

Description of drawings

Fig. 1 is for having the digital oscilloscope block diagram of waveform compression and demonstration according to the present invention.

Fig. 2 is for being used for the peak detection circuit block diagram of waveform compression and demonstration according to the present invention.

Fig. 3 is for being used for the low-pass filter block diagram of waveform compression and demonstration according to the present invention.

Fig. 4 is for being used for the first filter stage block diagram of the low-pass filter of Fig. 3 according to the present invention.

Fig. 5 is for being used for the second filter stage block diagram of the low-pass filter of Fig. 3 according to the present invention.

Fig. 6 is for being used for the 3rd filter stage block diagram of the low-pass filter of Fig. 3 according to the present invention.

Fig. 7 is for being used for the 4th filter stage block diagram of the low-pass filter of Fig. 3 according to the present invention.

Fig. 8 is for being used for the additional filter stages block diagram of the low-pass filter of Fig. 3 according to the present invention.

Fig. 9 is the alternative low-pass filter stage figure according to alternate figures 4-7 filter stage of the present invention.

Figure 10 is the planimetric map according to the Wave data graphic presentation of prior art.

Figure 11 is the planimetric map by the Wave data graphic presentation of compression of the present invention and displaying scheme processing.

Embodiment

Consult Fig. 1 now, the digital oscilloscope 10 shown in the figure has input port 12 and is used to receive electric signal.Electric signal from input port 12 is imported into droop amplifier 14, bandwidth limiter 16 and one or more A-D converter (ADC) 18.The digital signal that obtains with the data sampling form is imported into three different compression modules-conventional withdrawal device 20, low-pass filter 22 and peak detctor 24 in parallel.The data sampling that is extracted with low-pass filtering from withdrawal device 20 and low-pass filter 22 is stored in main obtaining in the storer 26, and detects data sampling from the peak value of peak detctor 24 and be stored in peak-data and obtain in the storer 28.Main storer 26 and the peak-data obtained obtained storer 28 and physically can be the part of same storer, and be well-known as affiliated field.And peak-data is obtained storer 28 comparable master on length and is obtained storer 26 and will lack.Can be from the data sampling that obtains

storer

26,28 via digital data bus 30 accesses for further handling.What be connected to digital data bus 30 is random-access memory (ram) 32, graphics engine 34 and display logic 36.The output of display logic 36 is presented on the display 38 for watching.All elements of digital oscilloscope are all under the control of processor 40.

Below explanation is divided into peak value detection, low-pass filtering and shows each joint.In context, each joint all is to describe at the digital oscilloscope 10 with 200MHz exemplary bandwidth and 2GS/ sampling rate second.Though what illustrate is the circuit that only is used for a passage, digital oscilloscope 10 can have nearly four passages.Each passage all had droop amplifier 14 and bandwidth limiter 16 before ADC 18.As a whole, ADC 18 samples from the adjusted simulating signal of bandwidth limiter 16 with the speed of 2GS/ second, and produces 8 bit digital data samplings.Data sampling is organized into 8 data stream, and they are with every 4ns timing watch (clocked).Data stream is delivered to withdrawal device 20, low-pass filter 22 and peak detctor 24.In the following discussion data byte be cited as S (n), S (n+1) ..., S (n+7), wherein S (n) samples at last, S (n+1) is in the sampling of 0.5ns before, the rest may be inferred.Data byte is that S (n+8) arrives S (n+15) among the former 4ns, and the rest may be inferred.

Peak detctor 24 is found out minimum and maximum value in a usual manner in the continuous time period, and these values are stored in peak-data obtain in the storer 28.Low-pass filter 22 is removed high fdrequency component, extraction and these values is stored in the master and obtains in the storer 26.Withdrawal device 20 can transmit all data samplings, maybe can extract them, and its resulting value also is stored in main obtaining in the storer 26.Though withdrawal device 20 and low-pass filter 22 are shown independent circuit, they can be designed to single unit, and applied low-pass filtering amount can not regulated from having to the required grade of maximum.And as implied above, the peak-data of advocating peace is obtained

storer

26,28 and can be combined.The content of obtaining

storer

26,28 is sent to RAM 32 via digital data bus 30, and peak value detect data can " mark " for background data for showing, and extract/the low-pass filtering data can " mark " supply demonstration for foreground data.Graphics engine 34 is also referred to as rasterizer, takes out the data sampling that is obtained from RAM 32, and produces the waveform picture, and it also is stored among the RAM.Display logic 36 is got the waveform picture from RAM 32, itself and other graphical information is merged, and resulting image is delivered to display 38.Processor 40, it can be one or more microprocessors, controls these operations.For the discussion purpose of this paper, other process of digital oscilloscope 10 for example triggers, carries out and measures and by interface data to be sent to printer or computing machine etc. all not shown, but is well-known to these processes of those skilled in the art.

Peak value detects data and generally has recently the lower sampling rate of data from withdrawal device 20 and low-pass filter 22.But graphics engine 34 is put the data that peak value detects data and low-pass filtering/extraction together, although different sampling rates is arranged.Peak value detects the information of obtaining in the storer 28, except the information that relevant peak-to-peak signal amplitude is provided, also can regard warning message as.It provides following warning:

Burst pulse takes place;

There is high-frequency signal;

The dynamic range of ADC 18 just is exceeded.

Peak value detects data and occurs in pairs.A time period, every pair has maximal value and minimum value.Certainly, when the time period is two samplings, do not compress.But for simplifying circuit design, the time interval that is used for the time period is preferably the multiple of 8 data samplings.

Fig. 2 illustrates in greater detail peak detctor 24.For helping to understand, allow the enough fast pipeline stages of operating of circuit not give and illustrate.Each peak detection system clock (4ns) has 88 place values to arrive peak detctor 24 from ADC 18.These values S (n) enters separately input register 42 to S (n+7) by clock.The output of input register 42 enters maximum/minimum circuit 44.Each maximum/minimum circuit 44 all has digital comparator 46, and it finds out two the greater in the input.The output control multiplexer 48 of comparer 46, multiplexer 48 is routed to higher value " max " output, and smaller value is routed to " min " output.Four maximum/minimum circuit 44 of first row are found out maximal value and minimum value in to input register 42 at each.Four maximum/minimum circuit 50 of secondary series are divided into two lower circuit only finding out peaked two upper circuit and only find out minimum value.Two maximum/minimum circuit 52 of the 3rd row have the peaked top maximum/minimum circuit of finding out all 8 input registers 42 and find out the bottom maximum/minimum circuit of the minimum value of all 8 input registers.

Two maximum/minimum circuit 54 of last row are found out maximal value and the minimum value in long-time section.Consult the maximum/minimum circuit 54 that last lists portion, beginning in the time period, multiplexer 53 is selected-128 input ends that are used to be input to the maximum/minimum circuit so that its by another input end from the 3rd row maximum/minimum circuit 52 to maximum value register 56.This causes maximum value register 56 to contain S (n) the maximum to S (n+7).On all cycles subsequently of peak detection system clock, last maximum/minimum circuit 54 will compare from the former maximal value of maximum value register 56 with from the new maximal value of the 3rd row maximum/minimum circuit 52, and select the greater to become the fresh content of maximum value register.Minimum value register 58 produces minimum value in each time period in a similar fashion, and the value of following multiplexer 55 initial selected 127 is used to be input to following that last maximum/minimum circuit 54.After the last peak detection system clock in the time interval, be written into peak-data from the maximal value of maximal value and

minimum value register

56,58 and minimum value and obtain in the storer 28.

Low-pass filter 22 is carried out low-pass filtering, extracts subsequently, the data sampling value is write become owner of then and is obtained storer 26.Low-pass filter 22 is designed to leach high frequency component signal minimizing aliasing before extraction, otherwise aliasing can make rebuild low frequency signal become impossible.Low-pass filter 24 also reduces other unwanted high fdrequency component of noise and signal.Low-pass filter 22 must with the data sampling true-time operation that arrives with the speed that is exceedingly fast.When the abundant over-sampling of simulating signal, extraction in a small amount can not cause the aliasing of useful signal component.For example when analog bandwidth be 200MHz, and sampling rate is 2GS/ during second, extracts with 2 and obtains the 1GS/ sampling rate of second.Do not need wave digital lowpass filter to prevent aliasing, because the analog bandwidth of 200MHz has significantly reduced the above component of signal of nyquist frequency of 500MHz.Owing to do not need the high-speed figure low-pass filter when extracting with 2 in this example, so low-pass filter 22 only can be designed to 4 or higher extraction speed work.

Low-pass filter 22 can design by the feature of observing simple low pass filter kind.These simple wave filters belong to the moving average filter class, wherein minority continuous sampling are sued for peace.These wave filters can be described-first weighting with a series of weightings and " take advantage of " with first sampling, and second weighting " is taken advantage of " with second sampling, and the rest may be inferred, and all product summations are obtained wave filter output.For example, the wave filter with two continuous sampling additions is (11).It is 2 that this wave filter gains when DC, but gain is zero when half sampling rate.More commonly, this wave filter is described as (0.5 0.5), gain is 1 when DC like this, but in order to remove multiplier, filter coefficient is restricted to " 1 ".All these simple filter have full gain when DC, and have zero gain in some frequency, and are listed as following table.

Filter coefficient	Zero-frequency (Nyquist relatively)
Filter coefficient	Zero-frequency (Nyquist relatively)	11	1
101		11	1
101		111	2/3
1001	1/3，1	111	2/3
1001	1/3，1	1111	1/2，1
10001	1/4，3/4	1111	1/2，1
10001	1/4，3/4	10101	1/3，2/3
11111	2/5，4/5	10101	1/3，2/3
11111	2/5，4/5	100001	1/5，3/5，1
111111	1/3，2/3，1	100001	1/5，3/5，1
111111	1/3，2/3，1	1000001	1/6，1/3，5/6
1001001	2/9，4/9，89	1000001	1/6，1/3，5/6
1001001	2/9，4/9，89	1010101	1/4，1/2，3/4
1111111	2/7，4/7，6/7	1010101	1/4，1/2，3/4
1111111	2/7，4/7，6/7	10000001	1/7，3/7，5/7，1
11111111	1/4，1/2，3/4，1	10000001	1/7，3/7，5/7，1

When the series connection of these wave filters is placed, just combine from the zero-frequency of each wave filter, that is: the wave filter that is constituted by placement that (111) are connected with (1001) has zero-frequency at 1/3,2/3 and 1 time Nyquist.By many these wave filters series connection are placed and selective filter so that on each unwanted frequency almost, all be zero, low-pass filter 22 just has required feature.And extract be also included within, to convert data to a data stream from 8 parallel data streams.So the practice of filtering and extraction is before extracting the frequency that is aliased into low frequency region to be removed.

Low-pass filter 22 generally slightly be illustrated in Fig. 3.(generally see E.B.Hogenauer " AnEconomic Class of Digital Filters for Decimation and Interpolation " IEEETransactions on Acoustics, Speech and Signal Processing, ASSP-29 (2): 155-162,1981).8 data stream or pipeline from ADC 18 enter first filter stage 60.The output of first filter stage is extracted with 2, and enters second filter stage 70.Signal extracts with 2 once more, has only two data streams to leave second filter stage 70 like this.Signal is then by third and fourth filter stage 80,90.Output terminal at the 4th filter stage 90, signal can be stored in main obtaining in the storer 26, or extract and enter additional filter stages 100 with 2, for example Matthew P.Donadio the cascade integrator-comb filter (CIC) described in the article " CIC Filter Introduction " ( Htt: //users.snip.net/～donadio/cic.pdf).Additional filter stages 100 allows filtering on wide frequency ranges.The output of additional filter stages 100 can further be extracted (120) with M.At last, multiplier 122 will be write data to become owner of and obtain storer 26 then in the gain-adjusted to 1 of DC.

First filter stage 60 is programmable, and can be (10000001) or (1001).As shown in Figure 4, enter separately input pipeline register 62 from 8 pipelines of ADC 18 with every 4ns timing watch, and be marked as S0 to S7.Previous S1, S3 and S5 data timing watch enter additional pipeline register 64, and become S9, S11 and S13 data.Data S0 is the oldest, and data S1 is 500ps sampling early, and data S9 is the 4.5ns sampling before S0, and the rest may be inferred.In the high frequency sampling pattern, multiplexer 66 is selected the tops input, make T0 be S0 and S3 with (68), promptly (1001) take advantage of (S0 S1 S2 S3).The gain that this wave filter has during at 1/3 Nyquist or 333MHz and at 1GHz is zero.At low frequency mode, multiplexer 66 is selected the bottoms input, make T0 be S0 and S7 with (68), promptly (10000001) take advantage of (S0 S1 S2 S3 S4 S5 S6 S7).The gain that this wave filter has when 1/7,3/7,5/7 and 1 Nyquist or 143MHz, 429MHz, 714MHz and 1GHz is zero.First filter stage 60 is carried out with 2 extractions, and uses all 8 inputs.

As shown in Figure 5, the data T0 in each data stream is to the T3 1ns of being separated by now, and nyquist frequency is 500MHz.At high frequency mode, second filter stage 70 is (11), and gain is zero when 500MHz.At low frequency mode, second filter stage 70 is (1001), and gain is zero when 167MHz and 500MHz.Input register 72 receives output T0 from first filter stage 60 to T3, and is labeled as U0 to U3.Signal U1 is imported into another register 74, to produce output U5.U0 sues for peace (78) at high frequency mode and U1 via multiplexer 76, or in low frequency mode and U3 summation (78), and U2 sues for peace (78) at high frequency mode and U3 via multiplexer 76, or in low frequency mode and U5 summation (78).Second filter stage 70 is carried out with 2 extractions, to produce data stream V0 and V1, uses all four input T0 to T3.

At the 3rd filter stage 80, shown in more detail Fig. 6, data V0 and the V1 2ns of being separated by now, and nyquist frequency is 250MHz.Data are imported into input register 81 and are labeled as W0 and W1, and they are imported into distributor 82 again and are labeled as W2 and W3.W2 is imported into the 3rd register 83, to produce W4.At high frequency mode, AND gate 88 is blocked and (86) W0+W3 and W1+W4, and the output X0 of output summer 89 and X1 reflection W1+W2 and W2+W3 and twices (85) (87), such the 3rd filter stage 80 is (22), gain is zero when 250MHz.By AND gate 89, obtain wave filter (1221) at low frequency mode and (86) W0+W3 and W1+W4, it is the combination of wave filter (11) and (111).The gain that present the 3rd filter stage 80 has when 167MHz and 250MHz is zero.Because extract, 333MHz is aliased into 167MHz, and is removed.

The 4th filter stage 90 is shown in Fig. 7, and input X0 and X1 are transfused to register 91 and receive among the figure, and are labeled as Y0 and Y1, and they are again by subsequently register 93 receptions, and is labeled as Y2 and Y3.Combination Y0+Y1, Y1+Y2 and Y2+Y3 addition 95.Y1+Y2's and become an output Z1, and at high frequency mode, AND gate 97 is blocked Y2+Y3, so Z0 output is and Y0+Y1.Every 4ns is sent to the main storer 26 that obtains from the data of Z0 and Z1, and the permission data are stored with the speed of 500MS/ second.When high frequency mode and front end were 200MHz, the bandwidth of all

filter stages

60,70,80,90 was approximately 80MHz.In the DC of each wave filter of high frequency mode gain is 32.By becoming owner of displacement Z0 and Z1 output when obtaining storer 26 in that Z0 and Z1 output is write, this gain conversions is gain 1.The maximum gain that surpasses Nyquist is approximately 0.01, so considerably less signal is by aliasing.At low frequency mode, signal Y2+Y3 is by AND gate 97, and with Y0+Y1 summation (99), to produce Z0 output, that is: Z0 be Y0, Y1, Y2 and Y3's and.In this pattern, the gain of the 3rd filter stage 80 is zero when 250MHz and 500MHz, and only uses Z0 output to be used for extracting with 2.Nyquist frequency becomes 125MHz.All

filter stages

60,70,80,90 are all when low frequency mode, and gain is zero when 125MHz, 143MHz, 167MHz, 250MHz, 333MHz, 375MHz, 429MHz and 500MHz.When front end was 200MHz, the bandwidth of these wave filters was approximately 36MHz.The DC gain of wave filter is 2*2*6*4 or 96, and this gain is corrected later.The maximum gain that surpasses Nyquist is approximately 0.0008 and takes advantage of the DC gain, so also be that considerably less signal is by aliasing.

As shown in Figure 8, additional cic filter level 100 has three moving average filters able to programme, is input with Z0.These mobile filter devices can not filtering just allow signal pass through, but or filtering to extremely low frequency.Every 4ns new data arrives input end, and add up in first register 102 (101).Though first register 102 can have 20 extra highest significant positions, it finally still overflows.The output of first register 102 enters first shifter circuit 103, and it can that is: when K1 is zero, not be shifted K1 position-K1 of data right shift in the scope of 0-20.By the input of the carry in the totalizer 104 of using the back output of first shift unit 103 is rounded off (round), totalizer 104 is the parts that comprise second totalizer of second register 105.The output of second totalizer 105 enters second shift unit 106 with second shifted divisor K2.The shift unit 109 that the output of second shift unit 106 enters the 3rd totalizer 107,108 and has shifted divisor K3.Three difference channels 110 are arranged behind three totalizers.When the every 4ns timing watch of totalizer, difference channel 110 is by extracting clock work, but i.e.: the every 8ns of their timing watchs, 12ns etc. or every 4ns, but every 4ns, 8ns, 12ns etc. get new value.Multiplexer 112 in each difference channel 110 is selected a tap along shift register 114.If first tap is selected, and the every 4ns of ifs circuit is timed clock, and that is then done in one of each difference channel 110 reverse former totalizer adds up, and data do not change by additional cic filter level 100.

From single totalizer back is single difference channel, shows the work of bright totalizer and difference channel best.When two circuit timing watch together, this is a pair of promptly as simple moving average filter work.The item number that multiplexer 112 in the difference channel 110 is selected in the moving average.When having selected first tap, select one moving average, and wave filter allow the input pass through, do not change.Can consider that totalizer and difference channel all use modular arithmetic, just overflowing in output adder 116 can not cause the error in the output like this.When having selected second tap, the output that difference channel produces now be two successive values adding up by totalizer 101-109 and.During in having selected 8 taps last, produce 8 moving average.But output is not extracted.Additional cic filter level 100 is correctly exported in each clock generating.

The branch clock rate of being on duty is a half of totalizer clock rate, and difference channel 110 being seen inputs always reflect two accumulated values, cause the length doubles of moving average sequence.This output is also extracted with 2.When the clock ratio was N, the length of moving average sequence multiply by N, and output is extracted with N.By three totalizer 101-109 series connection is placed, after connect three difference channels 110, output is by three moving average sequential filterings of three difference channels regulations.

Along with clock ratio N increases, output increases with the three cubed speed of N.The gain of each wave filter is that N multiply by the register number in the negative path.Because the series connection of three wave filters, the gain of three wave filters of therefore additional cic filter level 100 is A*B*C*N*N*N, and wherein A, B and C are the register numbers in the negative path of first, second and the 3rd difference channel 110.For making filter frequencies drop to about 1KHz, the value of N is approximately 100 ten thousand, and full gain needs last difference channel 110 to have and contains about 75 register about 1E20 like this.But this point can be avoided with three shifter circuits 103,106,109.Along with N increases, these shifter circuits are discarded least significant bit (LSB), allow last difference channel 110 to use 16 bit registers.Multiplexer 112 in three difference channels allows large-scale wave filter to be programmed.There is several combinations to merit attention.5, the wave filter that produces of 6 and 8 moving average can connect 4 extraction thereafter, only less than centesimal any high frequency component signal by aliasing.3,4 and 5,4,5 and 6, or the low-pass filter that also produces of 5,6 and 7 moving average.

Additional cic filter level 100 back are the programmable decimation devices 120 that extract with M.The value of M can promptly not extract from 1, to 4, promptly selects one in per four.After multiplier circuit 122 is followed and in the end extracted, and proofread and correct the gain error that adds up by all filter stage 60-100.Alternative is, raw data can be write to become owner of and obtain storer 26, and gain is then in graphics engine 34 and be corrected when data are presented to the user on display 38.

Though used four low-pass filter stage 60-90 in the low-pass filtering described in Fig. 3-7 before cic filter level 100, alternative low-pass filter 130 is shown in Fig. 9, its alternative preceding four low-pass filter stage.To being less than or equal to 8 extraction speed, alternative low-pass filter 130 filtering datas, but, then use cic filter level 100 to greater than 8 extraction speed.With in the past the same, extract speed at all, all frequencies that are higher than Nyquist all are suppressed to below one of percentage.88 samplings are loaded on each system clock, enter top register 132.The order of sampled data is: D0 is sampling the earliest, and D7 is the sampling of obtaining at last.Data enter 8 layers of filter element 134-148 then downwards.At each layer, keep another precision position.When pending data arrived the bottom of layer 134-148, data were filtered.As shown in the figure, 64 filter elements are arranged, each filter element F (n) is associated with original D (n) input sample.Each filter element 152 is worth addition (154) with two, and the result is placed in the register 156.A value comes near the element in its last layer, and another value is selected by multiplexer 158, with one in two values value as downward transmission.After the value of having selected downward transmission, after being doubled, data pass through filter element.The input of remaining multiplexer is from in one deck but the input of the filter element of more keeping right.In the filter block each layer can both be used the simple filter that constitutes by two.Multiplexer is selected same own addition of D (0) expression.Multiplexer selects D (1) expression to add the value of neighbor filter element; Multiplexer selects D (2) expression to add the value of the filter element of two samplings to the right, and the rest may be inferred.Such filter element produces many different wave filters.

Multiplexer 158 in arbitrary layer is all identical---and TapSelect0 control can be set to control all filter elements 152 of ground floor (layer 0) to select contiguous sampled value; TapSelect1 control can be set to control down all filter elements of one deck (layer 1) to select 5 data value to the right; Or the like.Because the edge of the extensible wave filter layer of some multiplexer taps, so there is the adjunct circuit (not shown) on practice filters layer right side.Every system clock, new data be by 8 positions of right shift, and also 8 positions of right shift of the content of the register in the filter element 152 156.These adjunct registers are positioned at the right side of filter block 130 in logic, do not give illustrating.Top register 132 also extends to the right.

In all cases, the output of low-pass filter 130 is results of 8 additions, and when when inserting 8 bit data on top, the data in the bottom are 16 bit data like this.It doesn't matter for the order of applied filter operations, as long as this layer can be supported the selection of wave filter implementation.Can be used to before extraction, prevent that one group of low-pass filter of aliasing is shown in following table:

Extract

TapSelectx	1	2	4	8
TapSelectx	1	2	4	8	TapSelect0	0	2	4	8
TapSelect1	0	2	3	6	TapSelect0	0	2	4	8
TapSelect1	0	2	3	6	TapSelect2	0	2	2	5
TapSelect3	0	1	2	4	TapSelect2	0	2	2	5
TapSelect3	0	1	2	4	TapSelect4	0	1	2	3
TapSelect5	0	1	1	2	TapSelect4	0	1	2	3

TapSelect6	0	1	1	2
TapSelect6	0	1	1	2	TapSelect7	0	0	1	1
Bandwidth (MHz)	200	160	80	40	TapSelect7	0	0	1	1

For example, can be used for constituting the low-pass filter that bandwidth is 80MHz with the wave filter shown in the 4 row extractions in the table.This wave filter before extracting with 4 in order to prevent aliasing.Multiplexer 158 can change two independent multiplexers into, and last multiplexer is selected D (0), D (1) and D (2), and multiplexer is selected D (0), D (5) and D (8) down, and the output of multiplexer enters totalizer 154.The filter element of this minimizing except that typical bank of filters delay is arranged, also has additional realization to postpone.

The afterbody of low-pass filter 130 is four and passes through/round off layer 160 by the/piece 162 that rounds off.For low resolution, 16 low-pass filtering data can be rounded to 8 precision, extract and be stored in the storer 12; For high resolving power, 16 bit data can be extracted and be stored in the storer.When data are not filtered, allow sampled datas to pass through, do not change by/the piece 162 of rounding off.When data are filtered but when unrounded, 16 inputs are routed to two 8 outputs.Filtered and when rounding off when data, the input control 0x0080 that rounds off is added to d0, and exports the result on y0.By having 8 outputs, reduce by half by/routing channel number that the piece 162 that rounds off uses in the piece subsequently.Only use some outputs of filter block 130 during the extraction, decide on decimation pattern and speed.

Main obtain extraction is arranged in the storer 26/low-pass filtering sampling and obtain in peak-data has in the storer 28 after peak value detects sampling, display logic 36 must present image directly perceived and beautiful for the user now on display 38.Shown waveform neither should the signal in original signal misleads the user by illustrating not, also should not hide the signal of interest component that exists in the original signal.If two versions show with full intensity that all then the peak value of waveform signal detection version trends towards making extraction/low-pass filtering version fuzzy.So, need contrast this two versions someway.

Show two versions and do not make any fuzzy a kind of method be, detect version (background version) with the draw peak value of waveform of normal mode, when drawing extraction/low-pass filtering version, be reduced near the brightness of drawn peak detection signal in the past of low pass signal filtering version (prospect version) then.Figure 10 illustrates the peak detection signal of drawing with normal mode, and Figure 11 illustrates the same signal with peak value detection and low pass version.The brightness that peak value detects version is lowered near the low-pass filtering version time, allows the low-pass filtering version of signal is observed.The brightness of peak value detection version increases with the distance of distance low-pass filtering version.Pixel count between reformed background of brightness and the prospect version can be from 1 or 2 pixel adjustment to defined maximum number.

Do not need to draw peak value with full brightness earlier and detect version, reduce brightness later on again.Display logic 36 can produce two versions of waveform simultaneously.Display logic 36 is determined the distance of each peak-data pixel apart from low-pass filter signal, and with the suitable intensity peak-data pixel of drawing.When reducing peak value and detect the brightness of version, consider last, current and next low-pass filter signal point, it is better that display 38 seems.Brightness is meant the different shades of gray scale, and wherein brighter shade is exactly from the outstanding shade of background.Different colours can be used to replace shades of gray, and can use the different saturation grade of color.

The another way that makes two versions form the contrast demonstration provides user's " intensity " control, and it can differently treat two versions.For example, user's control can be regulated the intensity of prospect version pixel then in first scope from zero intensity to the maximal value adjusting background version pixel of stipulating from the default level to the maximum saturation value in second scope.Like this, have only prospect version pixel to be seen, or background version pixel prospect version pixel adjustment intensity relatively, or last prospect version pixel relatively the background version pixel be enhanced so that the dirigibility that maximum is provided for the user aspect the real views of sampled signal waveform is being provided.

Certainly also can only use different colors or saturation gradation to realize contrast to two versions.The user provides the demonstration of sampled signal waveform to have right to choose for which kind of technology of use.

When ADC 18 arrived their scopes terminal, the peak value that these codes are placed into signal detected in the version.When finding limit code, the low-pass filtering version of signal may be inaccurate.This information can be shown, so that the user knows the low-pass filtering version of not trusting signal.The intensity of low-pass filtering version or color can change to show that it may be inaccurate.Intensity can be that signal be cannot see.And color or intensity that peak value detects version also can change.Alert message or icon also can be placed on the display.

Wave data is shown as frequency spectrum sometimes, during particularly by use Fast Fourier Transform (FFT) (FFT) algorithm.When frequency domain information is shown, only use the low-pass filtering version of signal.There is a pair of strategy to can be used to guarantee that shown information is accurately.First, whether the portions of the spectrum that can not reliably show or not, has no reason to show the frequency spectrum up to nyquist frequency that is: ,-or near nyquist frequency, do not have signal, because remove by filtering, or alias component is arranged near nyquist frequency owing to inadequate filtering.Aliasing signal in a small amount is always possible, because wave filter no all roses.By FFT is shown to be restricted to that stronger signal is only arranged, extremely Ruo aliasing signal just is hidden.Alternative is, the FFT display part that may contain aliasing signal indicates, like this signal of user in will this zone of careful attention.These zones can be by adding shade or by using different background colors to indicate to demonstration.

The low-pass filtering version of signal is storing the main further filtering afterwards in the storer 26 of obtaining into.Additional filtering can be finished by microprocessor or graphics engine 34.This filtering has further reduced bandwidth.In the process that reduces bandwidth, the degree that aliasing reduces is improved.By strengthening by the gain under some frequencies of former algorithm, bandwidth also can increase very little degree.Interpolation also can be used to increase sampling rate.The horizontal resolution of display 38 is to showing that high-frequency information is provided with the limit.The filtering once more before constituting waveform image of low-pass filtering data is to remove the high-frequency information that is higher than nyquist frequency by the setting of display horizontal resolution.This makes waveform image and record length irrelevant, because record length is the function of low pass filter cutoff frequency, and cutoff frequency is the function that extracts post-sampling speed, that is, and and the noisiness seen in waveform image otherwise be the function of record length.And high frequency noise has been eliminated from the low-pass filtering waveform, makes track more clear.

The alternatives of low-pass filtering is to use the waveform that produces by wave-average filtering.When carrying out wave-average filtering, oscillograph 10 obtains many waveforms, and shows the average waveform image then.The average waveform image obtains like this: average first point in each waveform, and constituting first point in the average waveform image, second point in average then each waveform, to constitute second point in the average waveform image, the rest may be inferred.In practice, wave-average filtering is often carried out in a continuous manner, does not need once to keep in storer many waveforms.A special advantage of wave-average filtering is that it trends towards removing and nonsynchronous all signals of trigger, for example noise.Because this noise reduces, average waveform usually looks and is very similar to the low-pass filter waveform.

Can also use envelope waveform to substitute peak detecting waveform.Envelope waveform forms by obtaining many waveforms, and that is done during with the execution wave-average filtering is identical.By finding out in all waveforms that obtain first maximal value and minimum value, find out first pair of point in the envelope waveform, by finding out in all waveforms second maximal value and minimum value, find out second pair of point in the envelope waveform, the rest may be inferred.Do not extract and produce single waveform and represent from all waveform data by obtaining many waveforms, average or envelope waveform is represented another kind of compressed format.The major advantage of wave-average filtering is, it reduced noise and with incoherent other signal of trigger, but do not reduce the signal bandwidth synchronous with trigger.

Wave-average filtering can be carried out on many low-pass filtering waveforms, and waveform envelope can be carried out on peak detecting waveform.Resulting waveform shows can be from following each acquisition:

Low-pass filtering waveform with peak detecting waveform;

Average waveform with envelope waveform;

From the average waveform of many low-pass filtering waveforms and from the envelope waveform of many peak detecting waveform.

Be not that every kind of signal all is benefited from above-mentioned compression and displaying scheme.For example video waveform is shown in imitation mode of being seen demonstration on analog oscilloscope usually.This preferably by use extracting but not low-pass filtering realize.Then, use brightness is shown as the data that the scheme of function that signal is in the time of specific voltage shows extraction.

Though do not conform with requiredly, low-qualityer wave filter for example at the HiRes wave filter described in the above-mentioned background, can be used to replace good low-pass filter to produce above-mentioned demonstration.This low-qualityer wave filter is easy to realize, and still significantly reduces noise and reduce some aliasings.

Therefore, the invention provides a kind of compression and display technique, it provides low-pass filtering waveform and peak detecting waveform from input signal, this has reduced noise and has eliminated most of frequency alias, and these two waveforms are shown simultaneously in the mode that does not make contained information fuzzy in each waveform.

Claims

1. the numeral of compression sampling data before storage is obtained system, comprises:

Be used for high sampling rate digital signal to produce the parts of a plurality of sampled data streams;

Be used for that parallelly compressed described a plurality of sampled data streams are with the parts of the compression sampling data stream that produces each compression scheme simultaneously with a plurality of compression schemes, at least one is the low-pass filtering scheme in the described compression scheme; And

Be used to store the parts of described compression sampling data stream as the sample data of obtaining of each compression scheme.

2. numeral as claimed in claim 1 is obtained system, also comprises:

Be used for from each compression scheme the sample data of obtaining generate the figure waveform image to produce the parts of respective waveforms image; And

Be used for showing the described respective waveforms image that overlaps each other and do not make the parts of the contained information fuzzy of any described waveform image.

3. numeral as claimed in claim 1 is obtained system, and wherein said compression member comprises:

Low-pass filter is as described low-pass filtering scheme, leaching high frequency component signal from described a plurality of sampled data streams before extracting, aliasing minimized, reduces noise and to reduce other unwanted high fdrequency component of described electric signal; And

Peak detctor, in parallel with described low-pass filter, to catch the high frequency incident in the described electric signal.

4. numeral as claimed in claim 3 is obtained system, and wherein said compression member also comprises withdrawal device, and described withdrawal device is in parallel with described low-pass filter and described peak detctor, is used for from the optionally discarded data sample of described a plurality of sampled data streams.

5. numeral as claimed in claim 2 is obtained system, and wherein said generation parts comprise the parts that are used for further compressing from the fetched data of described memory unit when generating described waveform image.

6. numeral as claimed in claim 5 is obtained system, and wherein said further compression member comprises and is used for from obtaining average a plurality of waveform continuously to produce the parts of described waveform image.

7. numeral as claimed in claim 5 is obtained system, and wherein said further compression member comprises and is used for from obtaining a plurality of waveforms of envelope continuously to produce the parts of described waveform image.

8. numeral as claimed in claim 3 is obtained system, and wherein said low-pass filter comprises the HiRes wave filter.

9. numeral as claimed in claim 3 is obtained system, and wherein said low-pass filter comprises a plurality of filter stages of series connection, and each filter stage has the simple low pass filter design, and the required extraction factor is provided.

10. numeral as claimed in claim 9 is obtained system, and wherein said a plurality of filter stages comprise:

Low-pass filter stage has described a plurality of sampled data stream as importing and provide output, and described output is extracted to produce single sampled data stream with the required extraction factor; And

Additional filter has described single sampled data stream from described low-pass filter stage as input, and with described compression sampling data stream with required low pass bandwidth as output.

11. numeral as claimed in claim 10 is obtained system, the output of wherein said low-pass filter stage comprises the output of selecting from the group of being made up of output of high frequency sampling pattern and the output of low-frequency sampling pattern.

12. numeral as claimed in claim 10 is obtained system, wherein said low-pass filter stage comprises:

Input register is used to receive described a plurality of sampled data stream; And

The multilayer filter element is connected with described input register, so that described a plurality of sampled data stream flows through described layer, supplies to extract subsequently with a plurality of sampled data streams that produce filtering.

13. numeral as claimed in claim 10 is obtained system, wherein said low-pass filter stage comprises a plurality of low-pass filter stage of series connection, first filter stage has described a plurality of sampled data stream as the filtering output of importing and provide extraction, filter stage subsequently has the filtering output conduct input from the described extraction of last filter stage, and last filter stage has described single sampled data stream as output.

14. numeral as claimed in claim 10 is obtained system, wherein said additional filter stages comprises the integrator-comb filter of cascade, integrator-the comb filter of described cascade has the moving average filter able to programme of a plurality of series connection, with described single sampled data stream as the input and with described compression sampling data stream as output.

15. numeral as claimed in claim 10 is obtained system, wherein said low-pass filter also comprises and is used for further extracting described compression sampling data stream to produce the parts of compression sampling data stream.

16. numeral as claimed in claim 15 is obtained system, wherein said low-pass filter comprises that also being used for before described compression sampling data stream is provided to described memory unit the gain-adjusted of described low-pass filter is 1 parts at DC the time.

17. the electric signal acquisition methods of compression sampling data before storage comprises following steps:

With the described electric signal of high sampling rate digitizing, so as each sampling clock cycle produce a plurality of sampled data streams;

Use a plurality of compression schemes parallelly compressed described a plurality of sampled data streams simultaneously, to produce the compression sampling data stream of each compression scheme, at least one is the low-pass filtering scheme in the described compression scheme; And

Store the obtain sample data of described compression sampling data stream as each compression scheme.

18. method as claimed in claim 17 also comprises following steps:

Obtain generation figure waveform image the sample data from the institute of each compression scheme, to produce the respective waveforms image; And

Show the described respective waveforms image that overlaps and do not make contained information fuzzy in any described waveform image.

19. method as claimed in claim 17, wherein said compression step comprises following steps:

As described low-pass filtering scheme, the described a plurality of sampled data streams of low-pass filtering are leaching high frequency component signal, aliasing minimized, reduces noise and to reduce other unwanted high fdrequency component of described electric signal before extracting; And

With described low-pass filtering concurrently peak value detect described a plurality of sampled data stream, to catch the high frequency incident in the described electric signal.

20. method as claimed in claim 19, wherein said compression step also comprises following steps: detect with described low-pass filtering and described peak value and extract described a plurality of sampled data stream concurrently, with optionally discarded data sample from described a plurality of sampled data streams.

21. method as claimed in claim 18, wherein said generation step are included in when generating described waveform image further compression from the step of the fetched data of described storing step.

22. method as claimed in claim 21, wherein said further compression step comprise from obtain continuously average a plurality of waveforms to produce the step of described waveform image.

23. method as claimed in claim 21, wherein said further compression step comprise from obtain continuously a plurality of waveforms of envelope to produce the step of described waveform image.

24. method as claimed in claim 19, wherein said low-pass filtering comprise the step of using the HiRes wave filter to carry out filtering.

25. method as claimed in claim 19, wherein said low-pass filtering comprise the step that a plurality of filter stages of use series connection carry out filtering, each filter stage has the simple low pass filter design, and the required extraction factor is provided.

26. method as claimed in claim 25, wherein said a plurality of filter stages comprise:

Low-pass filter stage has described a plurality of sampled data stream as importing and provide output, and described output is extracted with the required extraction factor, to produce single sampled data stream; And

27. method as claimed in claim 26, the output of wherein said low-pass filter comprise the output of selecting from the group of being made of output of high frequency sampling pattern and the output of low-frequency sampling pattern.

28. method as claimed in claim 26, wherein said low-pass filter stage comprises a plurality of low-pass filter stage of series connection, first filter stage has described a plurality of sampled data stream as the filtering output of importing and provide extraction, filter stage subsequently has the filtering output conduct input from the described extraction of last filter stage, and last filter stage has described single sampled data stream as output.

29. method as claimed in claim 26, wherein said additional filter stages comprises the integrator-comb filter of cascade, integrator-the comb filter of described cascade has the moving average filter able to programme of a plurality of series connection, with described single sampled data stream as the input and with described compression sampling data stream as output.

30. method as claimed in claim 26, wherein said low-pass filtering step comprise that also the described compression sampling data stream of further extraction is to produce the step of compression sampling data stream.

31. method as claimed in claim 30, wherein said low-pass filtering step also are included in described compression sampling data stream is provided to before the described memory unit the gain-adjusted of described low-pass filter to when the DC being 1 step.

32. the method for the overlapping waveform image that a demonstration is derived from the sampling electric signal may further comprise the steps:

Use peak value to detect compression scheme and compress described sampling electric signal, to produce the background version of described sampling electric signal;

Use extraction/low-pass filtering scheme to compress described sampling electric signal, to produce the prospect version of described sampling electric signal;

Show described prospect version with default intensity grade; And

Show described background version with variable intensity grade.

33. method as claimed in claim 32, wherein said background version step display comprises following steps: the described background version pixel that shows contiguous described prospect version pixel with the intensity that reduces, when described background version pixel is more and more far away apart from described prospect version pixel, increase the intensity of described background version pixel gradually.

34. method as claimed in claim 32, wherein said background version step display comprises following steps: use the different brightness used with showing described prospect version pixel to show described background version pixel, wherein brightness comprises gray scale, tone and/or saturation degree.

35. method as claimed in claim 32, wherein said background version step display comprises following steps: controlling from the zero first scope adjusted user who changes the intensity of described background version pixel to predetermined maximum.

36. method as claimed in claim 35, wherein said prospect version step display comprises following steps: in the described user's control of the second scope adjusted of the intensity that changes described prospect version pixel from the default value to the maximum saturation value.